RegexHelper.h 9.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126
  1. //-------------------------------------------------------------------------------------------------------
  2. // Copyright (C) Microsoft. All rights reserved.
  3. // Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
  4. //-------------------------------------------------------------------------------------------------------
  5. #pragma once
  6. namespace Js
  7. {
  8. struct RegexMatchState;
  9. class RegexHelper
  10. {
  11. static const int MinTrigramInputLength=250000;
  12. //
  13. // Dynamic compilation
  14. //
  15. static bool GetFlags(ScriptContext* scriptContext, __in_ecount(strLen) const wchar_t* str, CharCount strLen, UnifiedRegex::RegexFlags &flags);
  16. public:
  17. static UnifiedRegex::RegexPattern* CompileDynamic(ScriptContext *scriptContext, const wchar_t* psz, CharCount csz, const wchar_t* pszOpts, CharCount cszOpts, bool isLiteralSource);
  18. static UnifiedRegex::RegexPattern* CompileDynamic(ScriptContext *scriptContext, const wchar_t* psz, CharCount csz, UnifiedRegex::RegexFlags flags, bool isLiteralSource);
  19. private:
  20. static UnifiedRegex::RegexPattern* PrimCompileDynamic(ScriptContext *scriptContext, const wchar_t* psz, CharCount csz, const wchar_t* pszOpts, CharCount cszOpts, bool isLiteralSource);
  21. //
  22. // Primitives
  23. //
  24. public:
  25. static UnifiedRegex::GroupInfo SimpleMatch(ScriptContext * scriptContext, UnifiedRegex::RegexPattern * pattern, const wchar_t * inputStr, CharCount inputLength, CharCount offset);
  26. static Var NonMatchValue(ScriptContext* scriptContext, bool isGlobalCtor);
  27. static Var GetString(ScriptContext* scriptContext, JavascriptString* input, Var nonMatchValue, UnifiedRegex::GroupInfo group);
  28. static Var GetGroup(ScriptContext* scriptContext, UnifiedRegex::RegexPattern* pattern, JavascriptString* input, Var nonMatchValue, int groupId);
  29. private:
  30. static void PropagateLastMatch
  31. ( ScriptContext* scriptContext
  32. , bool isGlobal
  33. , bool isSticky
  34. , JavascriptRegExp* regularExpression
  35. , JavascriptString* lastInput
  36. , UnifiedRegex::GroupInfo lastSuccessfullMatch
  37. , UnifiedRegex::GroupInfo lastActualMatch
  38. , bool updateRegex
  39. , bool updateCtor
  40. , bool useSplitPattern = false );
  41. static void PropagateLastMatchToRegex
  42. ( ScriptContext* scriptContext
  43. , bool isGlobal
  44. , bool isSticky
  45. , JavascriptRegExp* regularExpression
  46. , UnifiedRegex::GroupInfo lastSuccessfullMatch
  47. , UnifiedRegex::GroupInfo lastActualMatch );
  48. static void PropagateLastMatchToCtor
  49. ( ScriptContext* scriptContext
  50. , JavascriptRegExp* regularExpression
  51. , JavascriptString* lastInput
  52. , UnifiedRegex::GroupInfo lastSuccessfullMatch
  53. , bool useSplitPattern );
  54. static bool GetInitialOffset(bool isGlobal, bool isSticky, JavascriptRegExp* regularExpression, CharCount inputLength, CharCount& offset);
  55. static JavascriptArray* CreateMatchResult(void *const stackAllocationPointer, ScriptContext* scriptContext, bool isGlobal, int numGroups, JavascriptString* input);
  56. static void FinalizeMatchResult(ScriptContext* scriptContext, bool isGlobal, JavascriptArray* arr, UnifiedRegex::GroupInfo match);
  57. static JavascriptArray* CreateExecResult(void *const stackAllocationPointer, ScriptContext* scriptContext, int numGroups, JavascriptString* input, UnifiedRegex::GroupInfo match);
  58. template<typename T> static T CheckCrossContextAndMarshalResult(T value, ScriptContext* targetContext);
  59. //
  60. // Regex entry points
  61. //
  62. public:
  63. static Var RegexMatchResultUsed(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  64. static Var RegexMatchResultUsedAndMayBeTemp(void *const stackAllocationPointer, ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  65. static Var RegexMatchResultNotUsed(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  66. static Var RegexMatch(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, bool noResult, void *const stackAllocationPointer = nullptr);
  67. static Var RegexMatchNoHistory(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, bool noResult);
  68. static Var RegexExecResultUsed(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  69. static Var RegexExecResultUsedAndMayBeTemp(void *const stackAllocationPointer, ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  70. static Var RegexExecResultNotUsed(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  71. static Var RegexExec(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, bool noResult, void *const stackAllocationPointer = nullptr);
  72. static BOOL RegexTest(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  73. template<bool mustMatchEntireInput> static BOOL RegexTest_NonScript(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, const wchar_t *const input, const CharCount inputLength);
  74. private:
  75. static void PrimBeginMatch(RegexMatchState& state, ScriptContext* scriptContext, UnifiedRegex::RegexPattern* pattern, const wchar_t* input, CharCount inputLength, bool alwaysNeedAlloc);
  76. static UnifiedRegex::GroupInfo PrimMatch(RegexMatchState& state, ScriptContext* scriptContext, UnifiedRegex::RegexPattern* pattern, CharCount inputLength, CharCount offset);
  77. static void PrimEndMatch(RegexMatchState& state, ScriptContext* scriptContext, UnifiedRegex::RegexPattern* pattern);
  78. static void ReplaceFormatString
  79. ( ScriptContext* scriptContext
  80. , UnifiedRegex::RegexPattern* pattern
  81. , JavascriptString* input
  82. , UnifiedRegex::GroupInfo match
  83. , JavascriptString* replace
  84. , int substitutions
  85. , __in_ecount(substitutions) CharCount* substitutionOffsets
  86. , CompoundString::Builder<64 * sizeof(void *) / sizeof(wchar_t)>& concatenated );
  87. public:
  88. static Var RegexReplaceResultUsed(ScriptContext* entryFunctionContext, JavascriptRegExp* regularExpression, JavascriptString* input, JavascriptString* replace);
  89. static Var RegexReplaceResultNotUsed(ScriptContext* entryFunctionContext, JavascriptRegExp* regularExpression, JavascriptString* input, JavascriptString* replace);
  90. static Var RegexReplace(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, JavascriptString* replace, JavascriptString* options, bool noResult);
  91. static Var RegexReplaceFunction(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, JavascriptFunction* replacefn, JavascriptString* options);
  92. static Var StringReplace(JavascriptString* regularExpression, JavascriptString* input, JavascriptString* replace);
  93. static Var StringReplace(JavascriptString* regularExpression, JavascriptString* input, JavascriptFunction* replacefn);
  94. static Var RegexSplitResultUsed(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, CharCount limit);
  95. static Var RegexSplitResultUsedAndMayBeTemp(void *const stackAllocationPointer, ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, CharCount limit);
  96. static Var RegexSplitResultNotUsed(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, CharCount limit);
  97. static Var RegexSplit(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, CharCount limit, bool noResult, void *const stackAllocationPointer = nullptr);
  98. static Var RegexSearch(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  99. static Var StringSplit(JavascriptString* regularExpression, JavascriptString* input, CharCount limit);
  100. static bool IsResultNotUsed(CallFlags flags);
  101. private:
  102. static void AppendSubString(ScriptContext* scriptContext, JavascriptArray* ary, CharCount& numElems, JavascriptString* input, CharCount startInclusize, CharCount endExclusive);
  103. template <bool updateHistory>
  104. static Var RegexMatchImpl(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, bool noResult, void *const stackAllocationPointer = nullptr);
  105. static Var RegexExecImpl(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, bool noResult, void *const stackAllocationPointer = nullptr);
  106. static Var RegexReplaceImpl(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, JavascriptString* replace, JavascriptString* options, bool noResult);
  107. static Var RegexReplaceImpl(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, JavascriptFunction* replacefn, JavascriptString* options);
  108. static Var RegexSearchImpl(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input);
  109. __inline static UnifiedRegex::RegexPattern *RegexHelper::GetSplitPattern(ScriptContext* scriptContext, JavascriptRegExp *regularExpression);
  110. static Var RegexSplitImpl(ScriptContext* scriptContext, JavascriptRegExp* regularExpression, JavascriptString* input, CharCount limit, bool noResult, void *const stackAllocationPointer = nullptr);
  111. static int GetReplaceSubstitutions(const wchar_t * const replaceStr, CharCount const replaceLength, ArenaAllocator * const tempAllocator, CharCount** const substitutionOffsetsOut);
  112. };
  113. }